CONSTRUCTION OF A SYNTHETIC SINGLE

DOMAIN ANTIBODY PHAGE DISPLAY

LIBRARY FOR MOLECULAR DIAGNOSTIC

APPLICATIONS

NUR HIDAYAH BINTI HAIRUL BAHARA

UNIVERSITI SAINS MALAYSIA

2015

CONSTRUCTION OF A SYNTHETIC SINGLE

DOMAIN ANTIBODY PHAGE DISPLAY

LIBRARY FOR MOLECULAR DIAGNOSTIC

APPLICATIONS

by

NUR HIDAYAH BINTI HAIRUL BAHARA

Thesis submitted in fulfillment of the requirements

for the Degree of Master of Science

August 2015

ii

ACKNOWLEDGEMENT

In the name of Allah SWT, the Most Gracious and the Most Merciful, I offer

my humble gratitude to You for giving me the strength to complete this thesis.

First and foremost, my deepest gratitude and sincere appreciation goes to my

dedicated supervisor, Dr. Lim Theam Soon whom has been the best mentor anyone

could possibly have. I thank him for his constructive ideas, criticism, guidance and

patience throughout the course of my study. I also would like to thank my co-

supervisor Dr. Choong Yee Siew. They have successfully guided me through some

difficult times and were always willing to sharpen my understanding of my research.

Most importantly, I am also greatly indebted to most beloved parents,

husband and my siblings for their endless love and support of me to achieve my

goals in life. My love for them transcends all boundaries. I could never have gotten

where I am today without their encouragement. They have never ceased to comfort

me in my darkest hours in completing my study and also many thanks go to my

siblings for their love and moral support. I feel blessed to have them in my life

My heartfelt appreciation and love goes to my beloved lab mates; Lim Bee

Nar, NoorSharmimi Omar, Chin Siang Tean, Loh Qiuting, Nur Faezee Ismail, Chin

Chai Fung, Anizah Rahumatullah and Chan Soo Kim .They have always stood by me

for four years through many phases of hardship and turbulence. I could not have

completed this study without their generous help.

Lastly, I would like to acknowledge the financial support received from The

Malaysian Ministry of Higher Education, My Brain and INFORMM, Universiti

Sains Malaysia.

iii

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS ii

TABLE OF CONTENTS iii

LIST OF TABLES viii

LIST OF FIGURES ix

LIST OF ABBREVIATIONS xi

LIST OF PUBLICATION xv

ABSTRAK xvi

ABSTRACT xvii

CHAPTER 1- INTRODUCTION 1

1.1 Phage Display Antibody Library 2

1.2 Antibody Format 6

1.3 Domain Antibody: Unique Biophysical Properties 8

1.4 Synthetic Antibody Technology 12

1.5 Synthetic Domain Antibody Generation 13

1.6 Antibody Selection by In Vitro Panning 15

1.6.1 Panning Via Conventional Method 17

(Immunotubes and Microtitre Plate)

1.6.2 Semi-Automated Panning 19

1.7 Statement of Problem 22

1.8 Research Objectives 24

iv

CHAPTER 2 - MATERIALS AND METHODS 25

2.1 Materials 25

2.1.1 Consumables 25

2.1.2 Kits 25

2.1.3 Equipment and Software 26

2.1.4 Media and Buffers 26

2.1.5 Polymerases, Restiction Enzyme, 29

Ligase and Phosphotase

2.1.6 Proteins Marker, DNA Marker, dNTPS 29

and Antibodies

2.1.7 Microorganisms and Eukaryotic cell lines 30

2.1.8 Plasmids and Antibiotics 30

2.1.9 Primers 31

2.1.9.1 VH Gene and Sequencing Primers 31

2.2 Methods 32

2.2.1 Antigen Preparation 32

2.2.1.1 Polymerase Chain Reaction(PCR) 32

2.2.1.2 Cloning 36

2.2.1.2.1 Digestion of PCR Amplified Genes By 36

Restriction endonucleases

2.2.1.2.2 Dephosphorylation of digested plasmids 36

2.2.1.2.3 Ligation of DNA Fragments 36

2.2.1.3 Protein Expression 40

2.2.1.4 Protein Purification 40

2.2.1.5 SDS-PAGE 40

v

2.2.1.6 Western Blot 41

2.2.2 Synthetic Single Domain Antibody Library Generation 42

2.2.2.1 Bioinformatic Analysis of Variable Heavy

Chain Genes 42

2.2.2.2 PCR Assembly of Variable Heavy Chain Genes 42

2.2.2.3 Cloning 43

2.2.2.3(a) Ligation Mix 43

2.2.2.3(b) Transformation protocol 43

2.2.2.4 Library Size Estimation and Calculation 44

2.2.2.5 Helper Phage Preparation 44

2.2.2.6 Phage Library Packaging 45

2.2.2.7 Phage Display Panning 46

2.2.2.7(a) Microtitre Plate Panning 46

2.2.2.7(b) Semi-Automated Panning Using 47

Magnetic Processor

2.2.2.8 Phage Polyclonal ELISA 50

2.2.2.9 Phage Monoclonal ELISA 50

2.2.2.10 Soluble Monoclonal Antibody Expression 52

2.2.2.11 Soluble Monoclonal Antibody ELISA 52

2.2.2.12 DNA Sequencing 53

CHAPTER 3 - RESULTS 54

3.1 Bioinformatics Analysis and Design of Heavy 54

Chain Variable Region Repertoire

3.2 Generation of Heavy Chain Variable Region Repertoire 55

3.3 Synthetic Domain Library Generation 62

vi

3.3.1 Diversity Analysis for The Complementary 65

Determining Region, CDR

3.4 Antigen preparation 68

3.4.1 Cloning of Recombinant Red and Yellow 68

Fluorescent Protein

3.4.2 Evaluation, Expressions and Purification of 72

Antigens.

3.5 Conventional Microtitre Plate Antibody Selection Process 78

3.5.1 Polyclonal ELISA Evaluation of Panning Rounds 78

3.5.2 Monoclonal ELISA Evaluation of Selected Clones 80

3.6 Semi- automated antibody selection process 82

3.6.1 Coupling of Biotinylated Antigens to Streptavidin Beads 82

3.6.2 Polyclonal ELISA Evaluation of Panning Rounds 85

3.6.3 Monoclonal ELISA Evaluation of Selected Clones 87

3.7 Monoclonal Antibody Evaluation 89

3.7.1 Monoclonal Antibody Cross Reactivity ELISA 89

3.7.2 Sequencing Analysis of Monoclonal Antibody 91

3.7.3 Monoclonal Antibody Solubility Expression ELISA 93

3.7.4 Western Blot Analysis of Monoclonal Antibody 96

Solubility Expression

CHAPTER 4 - DISCUSSION 98

4.1 Synthetic Single Domain Library Generation 98

4.2 Phage Display Selection 102

4.3 Synthetic Single Domain Antibody Library Selections

with Different Antigens 104

vii

4.4 Mtb 16 kDa Hsp Monoclonal Antibody Generation 106

4.5 Limitation of Study 108

4.6 Future Studies 110

CHAPTER 5 - CONCLUSION 111

REFERENCES 112

APPENDIX 119

viii

LIST OF TABLES

Page

2.2.1 Steps involved in PCR 33

2.2.2 PCR mixture 34

2.2.3 PCR condition 35

2.2.4 Ligation mixture of digested pRSET-BH6 and YFP 38

2.2.5 Ligation mixture of digested pRSET-BH6 and RFP 39

2.2.6 Automated Magnetic Bead-Based Panning Procedure 49

3.1 Average CDR distribution length 56

3.2 Chemically synthesized oligonucleotides primers, CDR, and 57

framework for VH genes

3.3 CDR diversity sequencing analysis of cloning colonies 66

3.4 List of antigens with their respective molecular weight 73

3.5 Sequencing clone analysis for the three CDR regions diversity of 92

antibody against 16 kDa antigen.

.

ix

LIST OF FIGURES

Page

1.1 Illustration of filamentous phage particle 5

1.2 Basic antibody Y shape structure 7

1.3 Schematic diagram of domain antibody format 11

1.4 Biopanning protocol 18

1.5 Semi-automated panning 21

3.1 Schematic diagram of VH gene design 59

3.2 Gradient PCR of VH gene assembly 60

3.3 PCR amplification of heavy chain 61

3.4 Colony PCR of library cloning transformation 64

3.5 Sequencing analysis of heavy variable domain 67

3.6 Red and Yellow fluorescent protein PCR amplification 70

3.7 Digested pRSET-BH6 plasmid 71

3.8 Image of fluorescent proteins under UV illumination at 302 nm 74

3.9 SDS Gel Analysis of yellow fluorescent protein purification 75

3.10 SDS Gel analysis of Ubiquitin protein purification 76

3.11 SDS Gel analysis of 16 kDA protein purification 77

3.12 Phage polyclonal ELISA of syn dAb against three FP 79

3.13 Phage monoclonal ELISA of syn dAb against three FP 81

3.14 Illustration of biotinylated protein conjugation on the 83

streptavidin bead

3.15 SDS Gel analysis of antigen conjugated with streptavidin 84

magnetic beads

x

3.16 Phage polyclonal ELISA of syn dAb against ubiquitin, 86

16 kDa

3.17 Phage monoclonal ELISA of syn dAb against ubiquitin, 88

16 kDa

3.18 Monoclonal cross reactivity ELISA of anti-16 kDa domain 90

antibody against eGFP.

3.19 Soluble syn dAb monoclones against 16 kDa protein 94

3.20 SDS Page analysis of soluble monoclonal

antibody against MTb 16 kDa Hsp antigen 95

3.21 Westernblot analysis of soluble syn dAb against 97

16 kDa protein

xi

LIST OF ABBREVIATIONS

Ab Antibody

ABTS 2,2’-azino-bis (3-ethylbenzthiazoline 6-sulfonic acid)

diamonium

Amp Ampicillin

ampR Ampicillin resistance gene (bla)

AVI-Tag Avidin-Tag

Bp Base pair

BSA Bovine serum albumin

Cam Chloramphenicol

CDR Complementarity determining region

Cfu Colony-forming unit

CH1 Constant Heavy Region 1

CH2 Constant Heavy Region 2

CL Constant Light

C-terminus Carboxy-terminus

Da Dalton

dAb Domain antibodi

ddH2O Double distilled water

DNA Deoxyribonucleic acid

dNTP’s Deoxyribonucleosid-5’-triphosphate

dsDNA Double stranded DNA

D segment Diversity segment

E. coli Escherichia coli

xii

EDTA Ethylendiaminotetraacetic acid

eGFP Enhanced green fluorescent protein

ELISA Enzyme-linked immunosorbent assay

Fab Fragment antigen binding

Fc Fragment crystalline

FP Fluorescent protein

Fv Variable fragment

Glu Glucose

GFP Green fluorescent protein

His-Tag Histidine Affinity Tag

hr Hour

HRP Horseradish peroxidase

Ig Immunoglobulin

IMAC Immobilised metal affinity chromatography

IPTG Isopropyl-ß-D-thiogalactoside

J segment Joining segment

Kan Kanamycin

Kb Kilo base pairs

kDa Kilo Dalton

KD Dissociation constant

M Mole / litre

mAb Monoclonal Antibody

MCS Multi-cloning site

Min Minutes

mL Millilitre

xiii

mm Millimeter

MP Milk powder

MPP Magnetic particle processor

MTP Microtitre plate

MW Molecular weight (in Dalton)

nmol Nanomoles per litre

Ni-NTA Nickel-nitrilotriacetic acid

OD Optical Density

OD600

OD at 600 nm wavelength

o/n Over night

PAGE Polyacrylamide gel-electrophoresis

PBS Phosphate buffered saline

PBS-T Phosphate buffered saline with Tween 20

PCR Polymerase chain reaction

PEG Polyethylene glycol

pfu Plaque-forming units

PP Polypropylene

PTM 2 % milk powder, 1 % Tween 20 in PBS

PVDF Polyvinylidene difluoride

RFP Red fluorescent protein

RNA Ribonucleic acid

rt Room temperature

rpm Revolutions per minute

scFv Single chain variable fragment

scFab Single chain fragment antigen binding

xiv

SDS Sodium dodecylsulfate

sec Seconds

SSB Single-strand DNA binding protein

ssDNA Single stranded DNA

syn dAb Synthetic domain antibody

Taq DNA polymerase from Thermus aquaticus

TEMED N,N,N’,N’-tetramethylethylenediamine

Tris Tris(hydroxymethyl)-aminomethane

Tween 20 Polyoxyethylenesorbitan monolaurate

U Enyzme units

Uv Ultra violet

V Volt

V-genes Variable genes

VH Variable domain of the immunoglobulin heavy chain

VL Variable domain of the immunoglobulin light chain

V segment Variable gene segment

YFP Yellow fluorescent protein

Mtb 16 kDa Hsp Mycobacterium tuberculosis 16 kDa heat shock protein

Units

(v/v) volume/volume

(w/v) weight/volume

μg Microgram

μL Microlitre

μm Micrometer

°C Degree Celsius

xv

LIST OF PUBLICATION

Hairul Bahara, N. H., G. J. Tye, et al. (2013). "Phage display antibodies for

diagnostic applications." Biologicals.

xvi

PENJANAAN PERPUSTAKAAN FAJ PAPARAN DOMAIN ANTIBODI

SINTETIK TUNGGAL UNTUK APLIKASI DIAGNOSTIK MOLEKUL

ABSTRAK

Antibodi domain telah dieksploitasi secara meluas sebagai perancah untuk

penjanaan perpustakaan antibody sintetik kerana saiz tanpa bergantung dengan

mekanisma lipatan mudah. Dalam kajian ini, penjanaan perpustakaan yang pelbagai

menggunakan rangka manusia tunggal (VH3-23(DP47)) dan kepelbagaian sintetik

diperkenalkan melalui kaedah mutasi rawak yang berlaku secara semula jadi dalam

kawasan kaset saling melengkapi (CDR), CDR1,CDR2 dan CDR3 pada rantaian

berat telah menghasilkan 10^9 saiz perpustakaan. Kepelbagaian sikuen bagi semua

CDR dapat ditentukan hasil daripada 28 klon yang dipilih secara rawak. Daripada 28

klon, 18 klon telah dipulihara dengan kepelbagaian panjang CDR3 yang berbeza dan

juga kepelbagaian dalam sisa amino asid. Kualiti perpustakkan yang dihasilkan dapat

dinilai melalui proses seleksi terhadap dua jenis antigen protein; penyakit dan protein

pendarfluor. Pelbagai klon sasaran unik khusus telah diperolehi bagi kebanyakan

antigen. Walau bagaimanapun, terdapat 2 antibodi monoclonal yang telah berjaya

diraih hasil daripada seleksi dengan Mycobacterium tuberculosis 16 kDa Hsp antigen

(Mtb 16 kDa Hsp) yang berpotensi untuk digunakan untuk tujuan teraputik.

Kesimpulannya, himpunan kepelbagaian perpustakaan naïf boleh digunakan pada

masa hadapan untuk menyaringi antibodi pengikat dengan antigen berpotensi yang

lain.

xvii

CONSTRUCTION OF A SYNTHETIC SINGLE DOMAIN ANTIBODY

PHAGE DISPLAY LIBRARY FOR MOLECULAR DIAGNOSTIC

APPLICATIONS

ABSTRACT

Domain antibodies have been widely exploited as a scaffold for the

generation of synthetic antibody libraries because of their relatively small size and

simple folding mechanism. In this study, the generation of a highly diverse library

using a single human framework (VH3-23(DP47)) and synthetic diversity introduced

by randomly mutating naturally occurring within complementarity-determining

regions (CDRs) CDR1,CDR2 and CDR3 of heavy chain yielded a library size of

10^9. The sequence diversity of all CDRs was determined from 28 randomly

selected clones. Out of the 28 clones, 18 clones were conserved with different length

of CDR3 and highly diverse amino acids residues. The quality of the library was also

validated by panning against two different types of protein antigens; diseases and

fluorescent proteins. Multiple unique target specific clones were obtained for most

antigens. However, two monoclonal antibodies were successfully raised against

Mycobacterium tuberculosis 16 kDa Hsp (Mtb 16 kDa Hsp) antigens which could

potentially be used for therapeutics. In conclusion, the diverse repertoire of the naïve

library can be used in the future to screen for binders against other potential antigens.

1

CHAPTER 1

1.0 Introduction

The rise of recombinant antibody technology was made possible by a

combination of innovations like polymerase chain reaction technology (Orlandi,

Güssow et al. 1989; Hoogenboom 2005), phage display technology (Siegel 2002)

and evolution of online data collection of human immunoglobulin genomic

sequences (Hust and Dübel 2004; Benhar 2007). For the past decade, the exploration

of an array of recombinant antibody libraries for various applications was carried out.

Improvements in molecular biology have paved the way towards improving

parameters in order to produce libraries with higher diversity, larger sizes and better

quality. Numerous studies have been done with the aspiration to mimic the uniquely

human adaptive immune system that constantly generates diverse binding capacities

of antibodies in a miniature sized test tube.

In 1975, the very first monoclonal antibody was introduced via hybridoma

technology that requires the immunization of animals (Muyldermans 2001).

Generation of these monoclonal antibodies involved the incorporation of myeloma

cells with antibody producing spleen cell (Kohler and Milstein 1975). Thus, the

hybrid will feature traits from both cells by maintaining immortality and antibody

production. Inevitably, the use of hybridoma technology to produce monoclonal

antibodies suffered several setbacks (Hoogenboom 2005). Some of the main

disadvantages of murine derived antibodies are the use of animal host, longer periods

of time required for production, unable to generate functional human antibodies and

incapable of generating antibodies against toxic antigens (Geyer, McCafferty et al.

2012). It is these bottlenecks that have made hybridoma technology an unattractive

2

prospect for antibody production. The degree of freedom on offer for researchers by

recombinant antibody technology has led to it gaining popularity in diagnostic

applications (Marks, Hoogenboom et al. 1991; Holt, Enever et al. 2000; Siegel 2002;

Ohara, Knappik et al. 2006).

In vitro display methods such as phage display, yeast display, ribosome

display technology were introduced as a major alternative for the generation of

recombinant human monoclonal antibodies (Barbas, Kang et al. 1991; Silacci, Brack

et al. 2005). It is an in vitro process that is independent of any regulation by the

immune system. The most widely used method is the phage display technology. This

method employs the use of filamentous bacteriophage M13 as the display machinery

(Barbas and Barbas 1994). The ability of a bacteriophage to present a recombinant

target on its surface was first evident with the pioneering work by George Smith with

peptides.

1.1 Phage Display Antibody Library

Phage display has earned its spotlight as the gold standard in vitro display

system that caters for the increasing demand for the generation of peptides and

recombinant proteins especially antibodies. The underlying concept of this display

technology is the physical linkage between genotype and phenotype. The robustness

of this technology lies in its ability to control and manipulate selection conditions. It

is therefore, independent of any regulation by the immune system. In addition,

antibodies can be harvested without having to go through animal immunization. Over

the last decade, in vitro display methods have been very successful in the generation

of diagnostic and therapeutic antibodies.

3

In general, there is an array of bacteriophages that has been exploited for

surface display such as T7, T4 and Lambda. However for phage display systems, the

most commonly used bacteriophage is the Ff class of filamentous phage. The Ff

phage comprises of M13, f1 and fd that belongs to the inoviridae family that infects

gram negative bacteria bearing the F-episome. It is a long rod like shape particle that

is made up of coat proteins encapsulating the single stranded genome. The viral coat

is mainly made up of 5 types of coat protein (pIII, pVIII, pIX, pVII, pVI). The

unique feature of the filamentous phage virus is the non-lytic lifecycle that has paved

the way for an in vitro tool to study the protein-protein interaction as well as

peptides. Phage propagation under the non –lytic cycle has allowed the phage display

system to function as a tool for surface display. In the early 1980s, George P. Smith

demonstrated the display of peptides via the fusion to the gIII gene of filamentous

phage surface. From this discovery, we are able to obtain information on the phage

physical linkage between genotype and its phenotype. The successful presentation of

peptides was achieved, the first phage derived antibody library for monoclonal

antibody production was reported (Winter, Griffiths et al. 1994)

Given the technological advancements over recent years, many researches

have attempted to display numerous proteins through fusion with different coat

proteins. However, with several limitations for display on each coat protein, only pIII

is vastly used to display large proteins. The major advantage of gIII fusion is that it

can tolerate relatively large insert without compromising the integrity of the F-pilus

infection. It’s worth mentioning that pVIII coat protein has also been used for display

of proteins and peptides. On the contrary, this fusion suffers from few drawbacks.

Because of the phage particles are vastly made up of pVIII coat protein, fusion of

large proteins or peptide for display may compromise the stability and structure of

4

the phage particles (Iannolo, Minenkova et al. 1995). Moreover, the fusion to gVIII

will correspond to avidity effects due to high copies of the protein being presented on

the surface, hence hindering affinity binding. In addition, the favored detection

system for M13 phage is based on antibodies to pVIII coat protein, therefore any

alteration to the gene VIII may interfere with the phage detection.

There are two ways for foreign proteins or peptides to be inserted as a fusion

to the phage coat proteins. It can be carried out using the phage vector or phagemid

vector system. In this study, the phagemid system with gene III fusion is employed

for the synthetic antibody library construction. Phagemids in general, are plasmids

with an existing E.coli plasmid origin of replication, multiple cloning sites and an

antibiotic-resistance gene inclusive of an additional Ff phage-derived origin of

replication and gene III or gene VIII. This addition allows for the phagemid to be

packaged as single stranded DNA (ssDNA) in viral particles. Phagemids can

function as normal plasmids or packaged as recombinant single stranded DNA in the

M13 capsid with the aid of a helper phage (Azzazy and Highsmith 2002). The added

advantage of using this phagemid system over phage vector is that soluble proteins

can be readily expressed in E.coli host without having to undergo any form of

alteration.

5

Figure 1.1 : Illustration of filamentous phage particle adapted from Eubanks 2007

(Eubanks, Dickerson et al. 2007). The phage particles are in linear rod like shape in

which consisting of single stranded DNA and five coat proteins. pVIII coat protein

also known as major coat protein makes up the vast structure of the phage protein

capsid. Fusion of foreign proteins and peptides are usually to the gene VIII and gene

III thus will be displayed fusion to pVIII and pIII coat protein.

6

1.2 Antibody Format

The classical format of antibodies are represented graphically as a Y-shape

structure (Figure 1.2) with two identical ends (Wood 2006). At N-terminus, a heavy

chain is linked via interchain disulphide bonds to a light chain to generate the

Fragment Antigen Binding (Fab) (Rader and Barbas 1997). The binding pockets of

the antibody is derived from the variable light and variable heavy domains within the

Fab structure (Huston, Margolies et al. 1996). The advancement of recombinant

antibodies through phage display has led to a wide array of different forms of

antibody formats to be introduced (Hudson 1998).

To date, formats such as the human domain antibodies, camelid domain

antibodies (Harmsen and De Haard 2007), single domain shark antibodies (Dooley,

Flajnik et al. 2003), single chain fragment variable (scFv), tandem scFv, diabody,

tetrabody, minibody and single chain fragment antigen binding have been

extensively employed as formats for monoclonal antibody generation (Andris-

Widhopf, Rader et al. 2000; Little, Kipriyanov et al. 2000; Holt, Herring et al. 2003;

Hussack, Keklikian et al. 2012). Moreover, the limitation introduced by the folding

machinery of Escherichia coli (Holliger and Hudson 2005) has resulted in the

preferred use of smaller fragments such as domain antibodies to be heavily utilized

for phage display (Holt, Herring et al. 2003; Dudgeon, Famm et al. 2009). Thus, the

introduction of the current formats is essential for researchers to curb such

limitations.

7

Figure 1.2 : Illustrations depicting the basic Y shape of an immunoglobulin and

smaller antibody derivatives commonly used for phage display, Fab (Fragment

antigen binding) and scFv (single chain fragment variable) adapted from Kierny

2012 (Kierny, Cunningham et al. 2012).

8

1.3 Domain Antibody: Unique Biophysical Properties

The first smallest known antigen-binding fragment coined as “nanobodies”,

‘‘domain antibody’’, or ‘‘dAb’’ was identified when a murine VH repertoire was

selected against the model antigen hen-egg lysozyme with high affinity and

specificity (Andris-Widhopf, Rader et al. 2000). Unlike scFv which is twice the size

and Fab, four times the size of dAbs (Holt, Herring et al. 2003), their relatively

smaller size is well suited for phage display (Chen, Zhu et al. 2009). Basically, dAb

is the variable regions of either the heavy (VH) or light chains (VL) of

immunoglobulins (Holt, Herring et al. 2003).

Recent commercial interest revolves around manufacturing humanized

antibody with high specificity and affinity for potential diagnostic and therapeutic

applications (Brekke and Sandlie 2003). The production of domain antibodies brings

about the advantages over the use of conventional antibodies. To add to its

commercial value, the antibody produced must meet the requirement of biophysical

properties (Harmsen and De Haard 2007) such as high yield and soluble expression

(Muyldermans 2001), heat stability (Goldman, Anderson et al. 2006) such as

resistance to proteolysis, resistance to harsh condition (Dona, Urrutia et al. 2009)

such as chemical degradation (Wang, Singh et al. 2007), aggregation and

denaturation.

Initial studies of domain antibody showed that the expression and solubility

of the first murine VH domain antibodies were low. The selection of the VH domain

was done in mouse with the presence of a cognate VL, therefore, it was thought that

the absence of the VH-VL hydrophobic interface contributed to the instability of the

structure. After the setbacks, a modification was introduced in cloning of camelid

9

VHH domains. It was found that the solubility improved due to a hydrophilic

mutation of a tetrad at the VL interface. Soon after, similar modifications of residues

at positions 44, 45 and 47 was done in human VH domains with those frequently

found in camel VHH domains. This approach is best known as ‘camelisation”

(Davies and Riechmann 1994; Conrath, Vincke et al. 2005). However, despite having

to overcome aggregation, the modified VH domains remained expressed at low

yields and relatively unstable due to the deformation of the ß-sheet.

In resolving this issue, many researchers studied the effect of the VHH tetrad

on solubility (Barthelemy, Raab et al. 2008) which resulted in a VH dAb library

produced based on a murine germline gene with a substitution at the VL interface.

Phage display panning was done with monomeric IgG-specific dAb and found to be

soluble at a concentration of 2 mM (Holt, Herring et al. 2003). On the same note,

good expression of fragments selected from the llama dAb library was attributed to

the framework substitutions that differs from the VHH tetrad. Mutation and

manipulation of the CDRH3 loop length also plays an important role to achieve good

expression and solubility of the VH antibody (Riechmann and Muyldermans 1999).

Besides having good expression yield and solubility, another attractive

property of several camelid VHH domains and llama VH domains is the heat

stability (Dudgeon, Famm et al. 2009). In general, antibodies and their fragments

derived from human VH dAbs tend to aggregate irreversibly upon heat denaturation.

However, it was reported that when camel and llama VHH domains (Dolk, van Vliet

et al. 2005) were subjected to heat ranging between 80–90⁰C, they were able to

maintain its antigen binding specificity despite prolonged incubation at high

temperatures. Advancements made to cater for the thermo stability includes site-

directed mutagenesis based approaches for directed evolution of antibodies.

10

Undoubtedly, successful isolation of recombinant antibodies from libraries

depend heavily in the quality of the libraries produced. Factors such as the number of

correctly folded functional antibodies have brought a paradigm shift towards

developing human domain antibody. An example of recent studies showed functional

antibody of HEL4 domain antibody library mimicking the natural human immune

response designed with only CDR3 diversity (Mandrup, Friis et al. 2013). This

library also includes mutations of the amino acid composition with regards to the

positions critical for the folding and aggregation of domain antibodies.

With regards to dAbs high affinity and specificity, their small size and short

half-life are best suited for targeting antigens in tissue and blood vessel where

penetration is often obstructed and for clearance purpose. For example in tumour

cells, dAb can be used to assist delivery of specific toxins to the tumour cells in a

short time without damaging healthy cells. However, in some applications, such as in

cancer treatment (Revets, De Baetselier et al. 2005) in which the target antigen

resides in the blood stream, prolonged serum half-life is crucial to maximize time for

antibody antigen reaction to minimizes the dosage amount.

11

Figure 1.3 : Schematic representation of antibody formats ranging from conventional

whole antibody to variable heavy and variable light chain domain antibody as the

smallest unit (Chakravarty, Goel et al. 2014).

12

1.4 Synthetic Antibody Technology

The rise of synthetic antibody stems from the limitation of natural repertoire

diversity. For the past twenty years, studies have shown a particular interest in

producing antibody with high affinity antigen-binding sites by introducing diversity

mutation in CDR loops. The construction of semi- or fully synthetic antibody library

genes were assembled using chemically synthesized DNA. One of the classical

methods of in-vitro antibody production derived from the natural antibody genes is

via PCR. Naïve antibody repertoires via synthetic platforms are not biased for

binders of any particular antigen and bypass the redundancies of naturally occurring

antibody. Hence, the advantage of synthetic antibody over the natural repertoire is

that it provides a wider diversity for any type of target.

Notably, technical advances have allowed the development of highly

functional synthetic antibody libraries that rival or even exceed the recognition

potential of natural immune systems. The first semi-synthetic library was reported in

1992 by Nissim and colleague with a human VH genes repertoire from 49 human

germline VH gene segments with combinatorial synthetic CDR3 of five or eight

residues (Hoogenboom and Winter 1992; Griffiths, Williams et al. 1994). Later on, it

was again followed up with a ‘single pot’ human scFv library built from a diverse

repertoire of in vitro human VH gene segments assembly with random nucleotide

sequence for CDR3 lengths between 4 – 12 residues (Benhar 2007).

The second generation of synthetic platform is based on a more limited

collection of variable domain genes however, emphasizing more on robustness. The

overall design took into consideration the yields of functional antibody fragments

based on cellular folding in the E. coli expression machinery (Welch, Govindarajan

et al. 2009). Since then, Pini’s group explored a semi synthetic antibody library that

13

was prepared using a single VH (DP47) and Vκ (DPK22)(Pini and Bracci 2000). The

VH component of the library was created using partially degenerate primers in a

PCR-based method to introduce random mutations at positions 95 – 98 in CDR3. It

was found that creating antibody libraries starting from well-expressed frameworks

was able to retain the diversity and stability (Hoogenboom, de ru ne et al. 1998).

The ‘HuCal’ libraries were constructed with a more diverse sequence space

although it is confined by the limited set of variable domain scaffolds (Benhar 2007).

All the genes assembled were synthetically synthesized with a total of seven VH and

VL (four Vκ and three Vλ) germline families that accounts for more than 95% of the

human antibody diversity (Knappik, Ge et al. 2000). In addition, the genes were also

optimized for expression in E. coli. The design of the library was based on cloning

the V genes of scFv in all 49 combinations into a phagemid vector. Diversity was

introduced in the CDR3 cassettes via generation of mixed trinucleotides sequences

by substitution CDR3 regions of the master genes. Interestingly, the outcome of the

library selection has resulted in obtaining high affinity binders with Kd between 10^9

M and 10^10 M. The variation of the CDR3 cassettes resulted in a highly diverse

library producing antibodies against a vast number of antigens with high affinity.

1.5 Synthetic Domain Antibody Generation

Interestingly, the unique nature of generating highly diverse antibodies

against a plethora of antigens by the immune system has intrigued researchers to

mimic such processes in vitro with synthetic gene platforms. The genetic sequence of

the variable domain is chemically synthesized with the introduction of randomization

at fixed positions corresponding to the CDR of the variable domain with a fixed

framework (Rothe, Urlinger et al. 2008; Yang, Kang et al. 2009; Prassler, Thiel et al.

14

2011). These degenerate oligonucleotides function as substitutes for the naturally

occurring in vivo diversity. Synthetic platforms also take into account the variation in

length of the CDR regions to fully maximize the diversity.

These oligonucleotides are designed using highly randomized codons which

are used to code for unspecified amino acids. The generation of amino acids

sequence depends on the codon usage. There are two commonly used codon scheme

of encoding unspecified amino acids sequences; 1) NNK ( A/T/C/G as an equimolar

representation of N and K codes for G/T ) 2) NNS ( N represents four bases and S

codes for G/C) (Barbas, Burton et al. 2001). These schemes produce 32 codons with

one stop codon. N in general, produces 64 possible codons, hence coding for 20

amino acids. The most commonly used is NNK as it is able to produce high

frequency of stop codon when used to encode for very large peptide consisting of

more than 50.

Construction of recombinant domain antibody requires the chemically

synthesized genes to be assembled in a manner that resembles the complete gene

sequence. The first method of assembly was introduced by Stemmer and colleagues

where full-length genes were generated (Stemmer, Crameri et al. 1995). This

approach is known as the conventional one-pot gene assembly (Prodromou and Pearl

1992; Stemmer, Crameri et al. 1995; Wu, Wolf et al. 2006). It is an annealing and

assembling process by incorporating the mixture of all synthetic oligonucleotide in a

single step PCR. However, due to the variation in the length of degeneracy, it is

rather difficult for gene assembly via the conventional method of polymerase chain

reaction. One-pot gene assembly is likely less efficient for degenerate

oligonucleotide with higher complexity as PCR is known to work well with a fixed

sequence and not randomized sequences (Young and Dong 2004). Confined by the

15

limitation, two-step approaches have been proposed to assist in the assembly of

genes with higher complexity for example by two-step PCR (Cherry,

Nieuwenhuijsen et al. 2008), ligation chain reaction of fragmented segments (Au,

Yang et al. 1998), gap filling and ligation (Ostermeier 2003).

Two-step PCR methods involve the assembly of multiple overlapping

oligonucleotides by PCR to generate the template DNA followed by the

amplification of the DNA template with two outermost oligonucleotides as primers.

The ligation chain reaction confers ligation of smaller fragments to form a unit and

subsequently amplified by PCR. The ligation chain reaction method however is

slightly similarl of the ligation chain reaction wherein, the genes are assembled by

polymerase gap filling in by ligating the ends together. Despite numerous proposed

approach proposed, it is worth noting however that these method are not routinely

used.

1.6 Antibody Selection by In Vitro Panning

Generation of antibodies by the immune system is involves the B cell

repertoire where the V genes segments have undergone rearrangement. As a result, a

single antibody is displayed on the surface of the each cell. The selection process

occurs by the interaction between antibodies with the antigen. Selected antibodies

will either segregate to short-lived plasma cells or to long-lived memory cells in

lymph nodes, spleen, and bone marrow (Winter, Griffiths et al. 1994). For memory

cells, the V genes of the selected antibodies will undergo hyper mutation. At this

point, binding affinity may be improved with successive selection with antigen. With

regards to mimicking the whole process of B cell antibody generation process,

“panning” or “biopanning” is used. iopanning refers to the iterative in vitro process

16

of antibody selection from antibody libraries based on target affinity (Kretzschmar

and von Rüden 2002).

There are several conditions that need to be taken into consideration during

the selection process. The first factor is the imperative proficiency of isolating the

gene pool to construct an antibody library with high diversity, capability of

expressing functional antibody fragment in soluble form and lastly, the efficiency of

simultaneous expression and display of genetic information being packaged. The

population of target specific antibodies are enriched relative to the number of

panning rounds (Mullen, Nair et al. 2006).

Target antigens are commonly coated on various solid phase. The most

common solid phase used are nitrocellulose, magnetic beads, agarose columns,

monolithic columns, polystyrene tubes and 96 well microtitre plates (Kontermann

and Dübel 2010). The solid phase bound phages are subjected to stringent washing to

eliminate nonspecific binders. The subsequent step is then followed by recovery of

the bound phages by elution. This process can either be by competitive elution

(Krishnaswamy, Kabir et al. 2009) or harsh acidic (Barbas, Kang et al. 1991) or

alkaline condition (Marks, Hoogenboom et al. 1991). Phage recovery or rescue plays

a pivotal role in the whole panning process as this will ensure retrieval of high

affinity binders. The phages are normally grown in bacterial culture for amplification

thus the recovered phages can be subjected to further rounds of selection. Moreover,

for each round of panning, phages can be enriched 20-1000 fold (McCafferty,

Griffiths et al. 1990).

17

1.6.1 Panning Via Conventional Method (Immunotubes and Microtitre Plate)

Prior to selection, the target antigens are coated on the surface of the solid

phase for presentation. Figure 4 shows the overall illustration of the conventional

panning process. This will be followed by an incubation step with the antibody

bearing phage particles to allow binding of antibodies to the antigen. Parameters such

as physical, chemical or biological are essentially introduced (Lee, Iorno et al.

2007). Stringent washing steps are necessary to ensure the removal of unbound

phage particles from bound phage particles. Discrepancy in the washing approach

will result in the variation of enriched clones. Lastly, an elution step can be

conducted in many ways either by enzymatic digestion, pH shift or competitive

antigen elution. The eluted phage particles are then enriched by infection of E. coli.

At this time, the phage particles can either be repackaged to be used in the

subsequent panning round or for final analysis.

After 4 to 6 rounds of panning, identification of bound phage can be

evaluated by antibody presenting phage or in the soluble form of antibodies on an

immunoassay format (Walter, Konthur et al. 2001). The positive clones will then be

sequenced to obtain the genotypic information pertaining to the positive clone. As

the genetic information of the clone is now available, modifications to the antibody

can be done and produced in different host depending on the platform the antibodies

will be used in. The availability of the genetic information of the antibodies would

also facilitate additional modifications in terms of stability and affinity maturation

(Pini and Bracci 2000).

18

Figure 1. 4 : Biopanning protocol adapted from Kügler 2013 (Kügler, Zantow et al.

2013). (a) Pooled phage library will be subjected to panning (b) Antigens are

immobilized onto solid surfaces and blocked to ensure nonspecific binding of phages

onto the plastic surface. (c) The pooled phages will then be incubated with the

immobilized antigen. (d) Unbound phages were then washed off by stringent

washing subsequently followed by elution. (e) Antigen bound phage will be rescue

by E.coli infection and followed by phage enrichment. (f) After every round of

enrichment, the phage can either be subjected to phage ELISA or carried forward

until the successive rounds completed.

19

1.6.2 Semi-Automated Panning

Screening large sample sizes is tedious with conventional panning procedures

that require repetitive rounds of panning, phage infection and propagation. However

in semi-automated systems to streamline the laborious process of phage display, such

as semi-automated magnetic bead-based antibody selection (Konthur, Wilde et al.

2010), allows high-throughput screening of antibodies to be carried out with

maximum convenience and minimal handling.

Conventional method of antigen immobilization using 96 well microtiter

plates involve two methods, either by adsorption of antigens to the plate surface

(Bora, Chugh et al. 2002) or coating the plates with streptavidin to capture the

antigen s(Välimaa, Pettersson et al. 2003). In contrast to using the microtiter plate,

another alternative is by allowing biotinylated antigens to be coated onto the

streptavidin magnetic beads (Cox and Ellington 2001). These magnetic beads have

larger surface area which contributes to the efficiency in the panning process as

compared to using microtiter plates.

In practice, the panning method utilizes a pin-based magnetic particle

processor (Kingfisher, Thermo) as shown in Figure 5(a). This machine enables the

handling of 96 magnetic pins in which it is positioned similar to the common 96 well

plate (Walter, Konthur et al. 2001; Rhyner, Konthur et al. 2003). The basic concept

of using the processor is to streamline processes such as washing step, incubation

times, and to conduct selections on same targets under different buffer conditions

simultaneously. The software-driven procedure dictates the transferring process of

magnetic particles between wells by capture and release motions shown in Figure

5(a). The rod-shaped magnets are covered with plastic caps during the transferring

process to prevent contamination. However, the automation process only involves

20

the panning procedure wherein the subsequent step of phage rescue and enrichment

is done manually. The advantage of semi-automated panning allows standardization

of panning parameters and reduces background of non-specific binder when

transferring from one well to another (Konthur and Walter 2002). The application of

this method allows better reproducibility and a faster turnover rate in comparison to

conventional plate based protocols. Therefore, the implementation of this method

allows for high-throughput antibody discovery.

21

Figure 1.5 : Semi-automated panning. (a) Diagram of King Fisher Flex machine

used to control automated beads panning. (b) Overall panning process from

incubation to washing and plate switching is done automatically. Figure adapted

from (Konthur, Wilde et al. 2010)

22

1.7 Statement of Problem

A vast number of commercially therapeutic antibodies approved by the U.S.

Food and Drug Administration are full-size antibodies of IgG1 format at about 150

kDa size (Holt, Herring et al. 2003). Due to their relatively large-size, these

molecules have limitations in terms of poor penetration into tissues (e.g., solid

tumors) and eventually results in weak binding to functionally important regions

(Dimitrov and Marks 2009). The use of using smaller formats can bring about

therapeutic relevance. As an example, the human immunodeficiency virus envelope

glycoprotein that can only be access by smaller molecules (Labrijn, Poignard et al.

2003). Therefore, by decreasing the size of the molecule it can aid tissue penetration

(Yokota, Milenic et al. 1993).

Over the last decade, a large amount of work has been focused on the

development of alternatives for smaller novel scaffolds. (Holt, Herring et al. 2003;

Holliger and Hudson 2005; Dimitrov and Marks 2009). Amongst the most explored

scaffold includes the relatively small domain antibody, which comprises of only the

domain antibodies and synthetic domain antibodies for various fields of research.

Most domain antibodies are derived from camelids, sharks and murine. This is

because fully human domain scaffolds of the variable gene repertoire are more likely

to aggregate. Since then, human heavy chain variable fragments (VH) have been

compared with those found naturally in camelids.

The determining factor for successful isolation of these antibodies relies

heavily on the quality of the library generated. Among the critical factors are based

on the diversity of the libraries as well as the functionality (Prassler, Thiel et al.

2011). While most of the studies conducted on the synthetic human domain antibody

tackles the issue of library construction (Silacci, Brack et al. 2005), biophysical

23

properties such as proper protein folding (Forrer, Jung et al. 1999) and aggregation

(Dudgeon, Famm et al. 2009) or diversity (Mondon, Dubreuil et al. 2008; Yang,

Kang et al. 2009). To circumvent this problem, mutational studies have been

conducted to understand the factors attributed to these problems.

The main focus for the antigen binding specificity lies within the CDR region

of the variable domain. In the early stages, in depth studies of domain antibody

sequence analysis have found that aggregation is likely to occur at the regions in or

adjacent to the CDR regions. Thus, the generation of synthetic domain antibodies

will allow for design of highly stable frameworks. Introducing diversity artificially

will eliminate any bias introduced by the host immune system. A full control of the

amino acid composition in the CDRs is possible by using the trinucleotide synthetic

design. Knappik et al pioneered a rather complex library by introducing diversity in

the CDR3 cassette in both variable heavy and light chain thus incorporating it in all

49 combination into phagemid vector (Knappik, Ge et al. 2000).

As more antibody sequence information was generated, several different

approaches have been proposed to improve diversity. Christ and his group developed

a synthetic human domain antibody library where the diversity was introduced in all

three of the CDR region (Lee, Iorno et al. 2007) in the heavy variable region to be

used in screening a wide array of antigen. In this study, the human domain antibody

constructed will be based on a known antibody framework that is reported to have

good solubility and stability (Lee, Liang et al. 2004; Mandrup, Friis et al. 2013).

Similarly, the method introduced in this study is aimed to focus on the assembly of

highly diverse genes of all three CDR regions with a defined single framework using

single-pot synthesis. The CDR lengths were determined via analysis of the average

length of CDRs naturally available.

24

The introduced method would help to establish a synthetic human domain

antibody library with unique and diverse CDR regions for functional antigen binding

by the extension of CDR-H2 and CDR-H3 distribution length. As more disease

specific biomarkers are being discovered, one of the major bottlenecks for the

development of diagnostic tests or even for basic research is the availability of

specific antibodies against these targets.

Therefore, this study has been conducted specifically for the production of

monoclonal antibodies against biomarkers with the use of a synthetic human domain

library. The naïve synthetic library will be used for selection of binders against a

wide range of disease specific recombinant antigen that can potentially be used in

diagnostic or even therapeutics platform.

1.8 Research Objectives

1. To design, assemble and clone a collection of synthetic human antibody variable

heavy genes

2. To generate a highly diverse in-house synthetic domain antibody phage display

library.

3. To identify monoclonal domain antibodies for potential binders against disease

specific recombinant antigen.